This papers explores important questions surrounding the sources of information marine predators incorporate into their foraging decisions. The questions are well addressed using an impressive dataset – the authors have clearly thought carefully about their predictions and how to apply a very clear and appropriate approach to test competing hypotheses. It’s also generally very well written, with the introduction and discussion in particular synthesising a large body of literature and forming a useful and compelling read; the methods and results could use some tightening up, and in places I was not available to fully evaluate the statistical methods due to some lack of clarity on terms and variables. Some aspects of the analyses may require revision, particularly decisions on how to structure models and whether to include random effects. Overall, I really enjoyed reading this!

Main comments:
1. Individual site fidelity terminology:
o Line 73: I would recommended here, and throughout the manuscript (e.g. line 301), using the standard term “individual site fidelity”, but “individual spatial fidelity” would also be better than “spatial individual fidelity” if the authors prefer to avoid the term “site”. Similarly line 74 “spatial individual specialization” should be “individual spatial specialization”.
o Line 78: Does individual fidelity itself fade, or does an individual’s probability of returning to a given site fade over time? An individual might return to its previous foraging site with a probability of 0.9; over time, the probability of it foraging at a different site increases, but its site fidelity of 0.9 remains the same: it just is now faithful to a different, newly-discovered site. I would suggest some rewording and clarification here.
o Line 93: “between individual divergence” is slightly unclear – perhaps “between individual difference”?
2. Lines 105-133: these questions are clear, but don’t explain why you explored effects of individual factors (e.g. sex, body mass, condition, etc.). Although you did not find an effect of these variables on consistency, it would be good to justify (briefly) why you looked for effects.
3. Line 125: were birds weighed once per season, or before (or after) each foraging trip? If birds were weighed only once per season, is there evidence this is unlikely to change much over the period that birds were being tracked? This is key to understanding whether effects of mass and body condition (line 184) are being accounted for.
4. Line 131: were the data regularised to account for different sampling frequencies? This feels relevant to how foraging trips were defined (line 136) as five GPS locations.
5. Lines 181-197: I found aspects of the statistical methods difficult to understand and to potentially require reconsideration. Specifically:
a) Line 184: I am slightly confused about the response variable here, as in line 170 it is explained to be a binary variable (<30° or not), but here it is referred to as a proportion. Perhaps for each individual you have calculated the proportion of pairs of trips that are <30°? If so, that should be added to the text.
b) Line 186: I don’t understand why including a fixed effect of body size measures means that a random effect of individual ID is not required here. If there are repeat measures per individual, this model should be rerun as a GLMM with a random effect of individual. Should year not also be included in this model?
c) In GLMM #2, I was confused as to why there are four different models fitted, each with different fixed effects, but with the same response variable. Why not fit a single model to compare all effects?
d) There appears to be no mention of model selection approach in the methods?
e) In general, references to the response variable in the models are confusing as multiple names appear to be used at different points in the text. E.g. “individual consistency in trip direction between pairs of consecutive trips” (line 187); “pairwise comparisons of annual individual consistency levels” (line 192). I recommend trying to refine and simplify this to improve the clarity of the paper, ideally by clearly naming and defining all of your response variables and using consistent names throughout.
f) Listing predictor variables in a single sentence would also make this section much clearer, e.g. lines 192-197: “Last, a GLMM was used to model the within individual consistency… including fixed effects of delta trip, year, time lag, and pairwise interaction between delta trip and year.”
6. In several places in the introduction and discussion (e.g. lines 324) the authors mention that the birds likely rely on social information. In the introduction, some further details on the types of social foraging strategies they use would be useful (e.g. local enhancement? Network foraging? Trail following?).

Minor comments:
7. Throughout the manuscript you use the term “average” – was this always the mean?
8. Line 89: s missing from “persists”
9. Line 139: unnecessary capitalisation of “gannets”
10. Line 140: the sentence here I found a bit confusing and some words were missing – I’d suggest reordering with the following sentence: “We summarised each foraging decision as the direction at maximum range from the colony: on average, 81.7% of locations recorded during a foraging trip fell within 30° of the direction at maximum range (stats), supporting this.”
11. Line 146: you state that the direction at maximum range has been mostly used previously – what has it been used for? Also “maximum” is repeated twice.
12. Line 147: “limitations” instead of “limits”.
13. Line 168: The brackets in GL(M)Ms, seems unnecessary – GLMMs is standard? This should be changed throughout the manuscript.
14. Lines 176-178: you mention that 30° also corresponds to the same order of magnitude as the width explored by birds in a trip, but this would be anything between 10-100° which doesn’t feel particularly similar? In any case, it likely makes sense to give the average angular width explored by birds during a single trip to back up this statement.
15. Line 225: “excluding 2020-21…” instead of “to the exclusion of”.
16. Line 245: This sentence is slightly unclear – what took 10.5 days?
17. Lines 266-268: Unclear what is meant by “yet with the majority of individuals above the average between similarity in direction”.
18. Line 327: “bird-borne” not “bird-born”
19. Line 342: “Alternatively,” instead of “Or else”
20. Figure 1: a key to indicate colours corresponding to depth categories would be much better than a written description here.
21. Figures 2-3:
o Figure titles shouldn’t be needed (e.g. in figure 2A remove “Angle between consecutive trips by same indiv.” – all information required to understand the figures should be in the axes titles and the legends.
o Axis titles for histograms should be “Angle between trip pairs (°)” or something similar rather than “bins of 10°”. Also ensure consistency of the terminology of your variables to the text, as per my comment 5e above.
o In Figure 2D the blue and black arrows are too similar to tell apart – can you use a larger arrow, or different colours?
o What is the black horizontal dashed line in figures 2C and 2D?

Comments
• I still think that the authors don’t properly justify why they look at individual traits (sex and body mass) on strategies in the introduction. The mention of “other traits” in line 58 isn’t very specific – as a minimum I would suggest at least adding (for example sex, body condition).

Minor points
• Line 15: replace “that” with “which” in “the Northern gannet that shows long-term…”
• Line 17: I think you should make clearer that the second clause is referring to within-individual behaviour: “… but at the within-individual level, birds persisted in following the same direction for longer timescales”
• Line 49: no need to capitalise “if”
• Line 62: “looping” suggests you are describing trip geometry here, whereas I think you simply mean they start and end at the same point (birds might make direct out-and-back trips that do not form a loop). I suggest using “central place” instead of “looping”.
• Figures 2 and 3 – The axis labels are very small and difficult to read, as well as the labels in the legend. I suggest using shorter labels and defining more fully in the legend so that the text is legible.
• Line 401 - <<>> symbols are not required around “ARS”.